All of lore.kernel.org
 help / color / mirror / Atom feed
From: Ingo Molnar <mingo@elte.hu>
To: "Chen, Kenneth W" <kenneth.w.chen@intel.com>
Cc: 'Nick Piggin' <nickpiggin@yahoo.com.au>,
	linux-kernel@vger.kernel.org, linux-ia64@vger.kernel.org,
	Andrew Morton <akpm@osdl.org>
Subject: [sched, patch] better wake-balancing
Date: Fri, 29 Jul 2005 14:13:11 +0000	[thread overview]
Message-ID: <20050729141311.GA4154@elte.hu> (raw)
In-Reply-To: <20050729114822.GA25249@elte.hu>


another approach would be the patch below, to do wakeup-balancing only 
if the wakeup CPU or the task CPU is idle.

I've measured half-loaded tbench and unless total wakeup-balancing 
removal it does not degrade with this patch applied, while fully loaded 
tbench and other workloads clearly improve.

Ken, could you give this one a try? (It's against the current scheduler 
queue in -mm, but also applies fine to current Linus trees.)

	Ingo

---

do wakeup-balancing only if the wakeup-CPU or the task-CPU is idle.

this prevents excessive wakeup-balancing while the system is highly
loaded, but helps spread out the workload on partly idle systems.

Signed-off-by: Ingo Molnar <mingo@elte.hu>

 kernel/sched.c |    6 ++++++
 1 files changed, 6 insertions(+)

Index: linux-sched-curr/kernel/sched.c
=================================--- linux-sched-curr.orig/kernel/sched.c
+++ linux-sched-curr/kernel/sched.c
@@ -1252,7 +1252,13 @@ static int try_to_wake_up(task_t *p, uns
 	if (unlikely(task_running(rq, p)))
 		goto out_activate;
 
+	/*
+	 * If neither this CPU, nor the previous CPU the task was
+	 * running on is idle then skip wakeup-balancing:
+	 */
 	new_cpu = cpu;
+	if (!idle_cpu(this_cpu) && !idle_cpu(cpu))
+		goto out_set_cpu;
 
 	schedstat_inc(rq, ttwu_cnt);
 	if (cpu = this_cpu) {

WARNING: multiple messages have this Message-ID (diff)
From: Ingo Molnar <mingo@elte.hu>
To: "Chen, Kenneth W" <kenneth.w.chen@intel.com>
Cc: "'Nick Piggin'" <nickpiggin@yahoo.com.au>,
	linux-kernel@vger.kernel.org, linux-ia64@vger.kernel.org,
	Andrew Morton <akpm@osdl.org>
Subject: [sched, patch] better wake-balancing
Date: Fri, 29 Jul 2005 16:13:11 +0200	[thread overview]
Message-ID: <20050729141311.GA4154@elte.hu> (raw)
In-Reply-To: <20050729114822.GA25249@elte.hu>


another approach would be the patch below, to do wakeup-balancing only 
if the wakeup CPU or the task CPU is idle.

I've measured half-loaded tbench and unless total wakeup-balancing 
removal it does not degrade with this patch applied, while fully loaded 
tbench and other workloads clearly improve.

Ken, could you give this one a try? (It's against the current scheduler 
queue in -mm, but also applies fine to current Linus trees.)

	Ingo

---

do wakeup-balancing only if the wakeup-CPU or the task-CPU is idle.

this prevents excessive wakeup-balancing while the system is highly
loaded, but helps spread out the workload on partly idle systems.

Signed-off-by: Ingo Molnar <mingo@elte.hu>

 kernel/sched.c |    6 ++++++
 1 files changed, 6 insertions(+)

Index: linux-sched-curr/kernel/sched.c
===================================================================
--- linux-sched-curr.orig/kernel/sched.c
+++ linux-sched-curr/kernel/sched.c
@@ -1252,7 +1252,13 @@ static int try_to_wake_up(task_t *p, uns
 	if (unlikely(task_running(rq, p)))
 		goto out_activate;
 
+	/*
+	 * If neither this CPU, nor the previous CPU the task was
+	 * running on is idle then skip wakeup-balancing:
+	 */
 	new_cpu = cpu;
+	if (!idle_cpu(this_cpu) && !idle_cpu(cpu))
+		goto out_set_cpu;
 
 	schedstat_inc(rq, ttwu_cnt);
 	if (cpu == this_cpu) {

  reply	other threads:[~2005-07-29 14:13 UTC|newest]

Thread overview: 60+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2005-07-28 23:08 Delete scheduler SD_WAKE_AFFINE and SD_WAKE_BALANCE flags Chen, Kenneth W
2005-07-28 23:08 ` Chen, Kenneth W
2005-07-28 23:34 ` Nick Piggin
2005-07-28 23:34   ` Nick Piggin
2005-07-28 23:48 ` Chen, Kenneth W
2005-07-28 23:48   ` Chen, Kenneth W
2005-07-29  1:25   ` Nick Piggin
2005-07-29  1:25     ` Nick Piggin
2005-07-29  1:39 ` Chen, Kenneth W
2005-07-29  1:39   ` Chen, Kenneth W
2005-07-29  1:46   ` Nick Piggin
2005-07-29  1:46     ` Nick Piggin
2005-07-29  1:53 ` Chen, Kenneth W
2005-07-29  1:53   ` Chen, Kenneth W
2005-07-29  2:01   ` Nick Piggin
2005-07-29  2:01     ` Nick Piggin
2005-07-29  6:27 ` Chen, Kenneth W
2005-07-29  6:27   ` Chen, Kenneth W
2005-07-29  8:48   ` Nick Piggin
2005-07-29  8:48     ` Nick Piggin
2005-07-29  8:53     ` Ingo Molnar
2005-07-29  8:53       ` Ingo Molnar
2005-07-29  8:59       ` Nick Piggin
2005-07-29  8:59         ` Nick Piggin
2005-07-29  9:01         ` Ingo Molnar
2005-07-29  9:01           ` Ingo Molnar
2005-07-29  9:07     ` Ingo Molnar
2005-07-29  9:07       ` Ingo Molnar
2005-07-29 16:40     ` Ingo Molnar
2005-07-29 16:40       ` Ingo Molnar
2005-07-29 11:48   ` [patch] remove wake-balancing Ingo Molnar
2005-07-29 11:48     ` Ingo Molnar
2005-07-29 14:13     ` Ingo Molnar [this message]
2005-07-29 14:13       ` [sched, patch] better wake-balancing Ingo Molnar
2005-07-29 15:02       ` [sched, patch] better wake-balancing, #2 Ingo Molnar
2005-07-29 15:02         ` Ingo Molnar
2005-07-29 16:21         ` [sched, patch] better wake-balancing, #3 Ingo Molnar
2005-07-29 16:21           ` Ingo Molnar
2005-07-30  0:08           ` Nick Piggin
2005-07-30  0:08             ` Nick Piggin
2005-07-30  7:19             ` Ingo Molnar
2005-07-30  7:19               ` Ingo Molnar
2005-07-31  1:15               ` Nick Piggin
2005-07-31  1:15                 ` Nick Piggin
2005-08-01 17:13                 ` Siddha, Suresh B
2005-08-01 17:13                   ` Siddha, Suresh B
2005-08-08 23:18           ` Chen, Kenneth W
2005-08-08 23:18             ` Chen, Kenneth W
2005-07-30 23:26         ` [sched, patch] better wake-balancing, #2 Chuck Ebbert
2005-07-30 23:26           ` Chuck Ebbert
2005-07-31  4:35           ` Con Kolivas
2005-07-31  4:35             ` Con Kolivas
2005-07-31  6:29           ` Ingo Molnar
2005-07-31  6:29             ` Ingo Molnar
2005-07-31 13:35         ` Chuck Ebbert
2005-07-31 13:35           ` Chuck Ebbert
2005-07-29 11:26 ` Delete scheduler SD_WAKE_AFFINE and SD_WAKE_BALANCE flags Ingo Molnar
2005-07-29 11:26   ` Ingo Molnar
2005-07-29 17:30 ` Chen, Kenneth W
2005-07-29 17:30   ` Chen, Kenneth W

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20050729141311.GA4154@elte.hu \
    --to=mingo@elte.hu \
    --cc=akpm@osdl.org \
    --cc=kenneth.w.chen@intel.com \
    --cc=linux-ia64@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=nickpiggin@yahoo.com.au \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.